Function strip_string

TransMLA_NeurIPS_2025/lighteval/math_utils.py:209–347

(string, skip_unit=False)

def strip_string(string, skip_unit=False):
    string = str(string).strip()
    # linebreaks
    string = string.replace("\n", "")

    # right "."
    string = string.rstrip(".")

    # remove inverse spaces
    # replace \\ with \
    string = string.replace("\\!", "")
    # string = string.replace("\\ ", "")
    # string = string.replace("\\\\", "\\")

    # matrix
    string = re.sub(r"\\begin\{array\}\{.*?\}", r"\\begin{pmatrix}", string)
    string = re.sub(r"\\end\{array\}", r"\\end{pmatrix}", string)
    string = string.replace("bmatrix", "pmatrix")

    # replace tfrac and dfrac with frac
    string = string.replace("tfrac", "frac")
    string = string.replace("dfrac", "frac")
    string = (
        string.replace("\\neq", "\\ne")
        .replace("\\leq", "\\le")
        .replace("\\geq", "\\ge")
    )

    # remove \left and \right
    string = string.replace("\\left", "")
    string = string.replace("\\right", "")
    string = string.replace("\\{", "{")
    string = string.replace("\\}", "}")

    # Remove unit: miles, dollars if after is not none
    _string = re.sub(r"\\text{.*?}$", "", string).strip()
    if _string != "" and _string != string:
        # print("Warning: unit not removed: '{}' -> '{}'".format(string, _string))
        string = _string

    if not skip_unit:
        # Remove unit: texts
        for _ in range(2):
            for unit_text in unit_texts:
                # use regex, the prefix should be either the start of the string or a non-alphanumeric character
                # the suffix should be either the end of the string or a non-alphanumeric character
                _string = re.sub(r"(^|\W)" + unit_text + r"($|\W)", r"\1\2", string)
                if _string != "":
                    string = _string

    # Remove circ (degrees)
    string = string.replace("^{\\circ}", "")
    string = string.replace("^\\circ", "")

    # remove dollar signs
    string = string.replace("\\$", "")
    string = string.replace("$", "")
    string = string.replace("\\(", "").replace("\\)", "")